AI Agent Algorithm Engineer Graduate (Global E-Commerce, Conversational AI) - 2027 Start Singapore Regular Bachelor/Master Graduate - 2027 Start
Core
Build and evolve the agent runtime (harness/agent loop) powering TikTok's global e-commerce customer service agents across 30+ languages.
Role type
Senior IC machine-learning engineer (AI agents)
Builds
Self-evolving agent runtime, context/memory systems, post-trained LLMs, and evaluation harnesses for customer service.
Domain
E-commerce, Conversational AI, Large Language Models
Deliverable
production ML models
Required skills
Python, C++/Go/Rust/Java, ML/DL/NLP fundamentals, LLMs/agents development, post-training (SFT/DPO/RL), agent systems (harness/context engineering/MCP), evaluation (LLM-as-judge), inference/serving
Preferred skills
Building agent runtimes/harnesses, RAG/tool-use agents, multilingual NLP, reward modeling, sub-agents, tool search
Technologies
vLLM, TensorRT-LLM, MoE, KV/prompt caching, MCP, SFT, DPO, RLHF, RLAIF, RLVR, τ-bench, SWE-bench, GAIA, BFCL
Responsibilities
Orchestrate skills, tools, and context; implement loop control and intervention; engineer context and memory for long multi-turn agents; post-train and fine-tune LLMs; design and integrate tools and MCP connectors; build evaluation harnesses; build the self-evolving loop with user simulators; own an end-to-end surface and ship to production.
Seniority
Junior, new grad
